Fix llama ONNX export #1432
Merged
Fix llama ONNX export #1432
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
echarlaix
reviewed
Oct 6, 2023
| @@ -216,7 +216,48 @@ class OPTOnnxConfig(TextDecoderOnnxConfig): | |||
| NORMALIZED_CONFIG_CLASS = NormalizedTextConfig | |||
|
|
|||
|
|
|||
| class LlamaDummyPastKeyValuesGenerator(DummyPastKeyValuesGenerator): | |||
There was a problem hiding this comment.
Would make sense to move it in optimum/utils/input_generators.py
optimum/optimum/utils/input_generators.py
Line 830 in 099cd73
| random_sequence_length_range=random_sequence_length_range, | ||
| **kwargs, | ||
| ) | ||
| self.num_key_value_heads = normalized_config.num_key_value_heads |
There was a problem hiding this comment.
should we also add a fix in prepare_inputs_for_merged
(as done in #1425)
There was a problem hiding this comment.
yes thank you for taking care of it!
echarlaix
added a commit
that referenced
this pull request
Oct 9, 2023
* Add ONNX export Mistral models support * add test * format * fix format * fix key _config * tmp install transformers from source for tests * change model id * fix after #1432 merged * fix * format * fix
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
LLama uses an optional configuration key
num_key_value_headsfor the number of heads, and usesnum_attention_headsto compute the head dimension. This was unfortunately not implemented in #975 (apart from llama2 70b, the llama and llama2 series do not make use of this num_key_value_heads key), as the key probably did not exist at the time.Fixes #1399